On 2007-12-09 17:28, meska_jd wrote:
> Ok, and now you came back to my initial point. if /b/ was [w] in IE
> (as they seem to suggest) and we have Celtic evidence that /p/ was
> very likely [P] (bilabial fricative) in the first place, then why
> can't we suppose that there were no [b] and [p] in IE at all, just [w]
> and [P]?
We can. The question is how much we gain by doing so. The pair would be
isolated in the phonological system and the game may not be worth the
candle if we solve the "mystery" of *p lenition in Celtic only to
complicate the straightforward development of *p > p almost everywhere
else (*p displays no strange behaviour outside Celtic). Perhaps *p was
simply particularly prone to aspiration?
PIE *w sometimes patterns with stops, as in the *wr-/wl- clusters, but
on the other hand it also behaves like an ordinary semivowel;
phonotactically, it doesn't pattern with *p in roots like //bHewg-//).
There are several mysterous roots with a final *w after a liquid or
laryngeal, e.g. *terh2w- 'overcome, cross over' (Hitt. tarhuzzi, OInd.
tú:rvati, tarute), and one might speculate that this final *w derives
from pre-PIE **b (a stop could well occur in this position). This would
mean that there were originally both **b and **w, and the former mostly
merged with the latter. It's just a loose suggestion, don't take it too
seriously.
Piotr